JP2018183601A - Diagnostic support system, operation method of the same, diagnostic support device, and program - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a diagnostic support system which can efficiently present information required to improve the certainty of diagnosis by a medical doctor.SOLUTION: A diagnostic support system 100 has: an input information acquisition section 102 which acquires input information such as an abnormal shade relating to a case being a target of medical diagnosis; an inference section 104 which identifies a diagnostic name corresponding to the case on the basis of the input information and acquires an inference probability of the diagnostic name; and a presentation section 110 which allows support information corresponding to the inference probability of the identified diagnostic name to be displayed.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a diagnosis support system, and more particularly to a diagnosis support system that provides information for supporting medical diagnosis.

  In the medical field, image diagnosis is performed in which a doctor interprets a medical image obtained by an imaging apparatus such as an X-ray CT apparatus or an MRI and makes a diagnosis. For example, in image diagnosis, a doctor who has received a request for interpretation by an attending physician can comprehensively determine the findings (hereinafter referred to as “interpretation findings”) and various measured values obtained from the images and determine the symptoms of the lesions appearing in the images. Identify. Then, the doctor summarizes the background of the image diagnosis as an interpretation report to the requesting doctor by using interpretation findings and measurement values.

  For the purpose of reducing the burden on doctors' interpretation, a medical diagnosis support apparatus (hereinafter referred to as “diagnosis”) detects abnormal shadows in medical images and infers the state of the shadows by computer processing. Has been developed. Usually, when using such a diagnosis support device, an interpretation is first performed by a doctor, and then the doctor refers to the support information provided by the diagnosis support device and compares it with the result of the interpretation by itself. Make a good judgment. Here, in order to trust the support information presented by the diagnosis support apparatus, it is desirable for the doctor to know what the support information was derived from. That is, it is preferable that the diagnosis support apparatus presents the reasoning reason for the support information to be presented.

  Therefore, a diagnosis support apparatus for presenting such an inference basis has been developed. For example, in Patent Document 1 proposed by the applicants of the present application, there is a technique for presenting negative information and positive information as support information for an inference result of an apparatus based on input information such as interpretation findings. Have been described. Further, Patent Document 1 describes a technique for presenting negative information and positive information for each possible diagnosis name as support information. Accordingly, the diagnosis support apparatus presents the inference result, and information that has influenced the derivation of the inference result of the apparatus based on information such as interpretation findings can be presented as support information.

Japanese Unexamined Patent Publication No. 2010-200840

  As a tendency of interpretation by doctors, information that affirms the assumed diagnosis name when there is confidence in the doctor's assumed diagnosis name, and information that denies other diagnosis names when there is no confidence Often used as a clue. However, the technique described in the above-described patent document has a problem that such information may not be presented as support information.

  In view of the above-described problems, an object of the present invention is to enable efficient presentation of information necessary for improving the certainty of diagnosis by a doctor.

  As a means for achieving the above object, the diagnosis support system of the present invention comprises the following arrangement. That is, an acquisition means for acquiring input information corresponding to a case, an inference means for specifying a diagnosis name corresponding to the case based on the input information and obtaining an inference probability of the diagnosis name, and the specified diagnosis name Display control means for displaying on the display unit support information corresponding to the inference probability.

  ADVANTAGE OF THE INVENTION According to this invention, information required in order to improve the reliability of a doctor's diagnosis can be shown efficiently.

The figure which shows the apparatus structure of the diagnosis assistance apparatus by 1st Embodiment. The figure which shows the basic composition of the computer which implement | achieves each part of a diagnostic assistance apparatus with software. The flowchart which shows the process sequence in 1st Embodiment. The figure which showed the example of the interpretation findings and clinical information in 1st Embodiment. The figure which showed the example of the influence degree in 1st Embodiment. The figure which showed the example of the assistance information in 1st Embodiment.

  Hereinafter, embodiments (embodiments) for carrying out the present invention will be described with reference to the drawings. However, the scope of the invention is not limited to the illustrated examples.

[First Embodiment]
The diagnosis support system according to the first embodiment acquires medical information about a case that is a medical diagnosis target, and performs diagnosis support for the case. In the present embodiment, the diagnosis support system acquires a plurality of interpretation findings, past medical histories, and tumor marker values (hereinafter referred to as clinical information) related to abnormal lung shadows as medical information. Then, based on the acquired information, an explanation will be given taking as an example a case where an inference regarding the type of abnormality (diagnosis name) of an abnormal shadow of the lung is performed and diagnosis support information is presented based on the result. Note that the inference target is not limited to this, and the diagnosis names, interpretation findings that can be input, clinical information, and the like described below are merely examples for explaining the processing steps of the diagnosis support system.

  FIG. 1 shows the configuration of the diagnosis support apparatus of the system according to the first embodiment. The diagnosis support apparatus 100 according to the present embodiment includes an input information acquisition unit 102, an inference unit 104, an influence level acquisition unit 106, a selection unit 108, and a presentation unit 110. Further, the diagnosis support apparatus 100 is connected to the information input terminal 200.

  The information input terminal 200 is a server (not shown) that displays case information (such as medical image and electronic medical record information including clinical information) related to an abnormal shadow of the lung corresponding to a case to be diagnosed by a user (doctor) operation or the like. Get from. The information input terminal 200 may be connected to an external storage device such as an FDD, HDD, CD drive, DVD drive, MO drive, ZIP drive, etc., and may acquire case information from these drives. The information input terminal 200 displays these case information in a form that allows interpretation.

  The user inputs interpretation findings for case information displayed on a monitor as a display unit using a mouse or a keyboard. This process is realized by the information input terminal 200 having a function that can be selected by the GUI using, for example, a template-type input support method. In the present embodiment, it is assumed that the interpretation findings and clinical information each include a name and a state, as will be described later. Further, in the information input terminal, the user may input the interpretation findings with reference to the monitor display, and the clinical information may be automatically input to the template, or may be input by the user. The information input terminal 200 transmits interpretation findings, clinical information, and data (such as representative images) associated with this information to the diagnosis support apparatus 100 via a LAN or the like in accordance with a user operation.

  The input information acquisition unit 102 includes information on abnormal lung shadows (hereinafter referred to as input information) input from the information input terminal 200 to the diagnosis support apparatus 100 and accompanying data (hereinafter referred to as input information and medical information). Are output to the inference unit 104, the influence level acquisition unit 106, and the presentation unit 110. Based on the input information acquired by the input information acquisition unit 102, the inference unit 104 performs inference processing related to the abnormal shadow of the lung to be diagnosed, and calculates the probability (inference result) that the abnormal shadow is each diagnosis name. To do. The calculated inference result is output to the influence degree acquisition unit 106, the selection unit 108, and the presentation unit 110. The inference process will be described later.

  The influence level acquisition unit 106 acquires the influence level using the input information acquired by the input information acquisition unit 102 and the inference result acquired by the inference unit 104, and outputs the acquired influence level to the selection unit 108. The method for acquiring the influence level will be described later. The selection unit 108 selects information based on the inference result acquired by the inference unit 104 and the influence degree acquired by the influence degree acquisition unit 106. The information selected by the selection unit 108 is output to the presentation unit 110. The presentation unit 110 generates and displays information to be presented. Specifically, the presentation unit 110 presents information based on the input information acquired by the input information acquisition unit 102 and accompanying data, the inference result acquired by the inference unit 104, and the information selected by the selection unit 108. And control the display of the generated information.

  Note that at least a part of each part of the diagnosis support apparatus 100 shown in FIG. 1 may be realized as an independent apparatus. Moreover, you may implement | achieve as software which each implement | achieves a function. In the present embodiment, each unit is assumed to be realized by software.

  FIG. 2 is a diagram illustrating a basic configuration of a computer for realizing the functions of the respective units illustrated in FIG. 1 by executing software. The CPU 1001 mainly controls the operation of each component. The main memory 1002 stores a control program executed by the CPU 1001 or provides a work area when the CPU 1001 executes the program. The magnetic disk 1003 stores an operating system (OS), device drives of peripheral devices, various application software including programs for performing processing described later, and the like. The display memory 1004 temporarily stores display data. The monitor 1005 is, for example, a CRT monitor or a liquid crystal monitor, and displays an image, text, or the like based on data from the display memory 1004. A mouse 1006 and a keyboard 1007 perform pointing input and character input by the user, respectively. The above components are connected to each other via a common bus 1008 so that they can communicate with each other.

  FIG. 3 is a flowchart showing a processing procedure in the present embodiment. In this embodiment, it is realized by the CPU 1001 executing a program that realizes the function of each unit stored in the main memory 1002. In the following description, the interpretation findings or clinical information names are represented by Ij (j = 1 to m), and m types of interpretation findings or clinical information names I1 to Im are handled. Also, k states that Ij can take are denoted as Sjk. The range of k varies depending on Ij. In the present embodiment, as an example, interpretation findings or clinical information as shown in FIG. 4 can be input or acquired, and each interpretation finding or clinical information takes a state as shown in FIG. It shall be possible. For example, the “shape” of I1 represents the shape of an abnormal shadow, and takes three states: state S11 “spherical”, state S12 “branch”, and state S13 “irregular”. The “cut” in I2 represents the degree of cut in the abnormal shadow. In addition, Il “roll (bronchi)” indicates the presence or absence of bronchial involvement in an abnormal shadow. In addition, the “history” of Im indicates whether or not there has been a history of the disease in the past.

  In the following description, a set of states Sjk is denoted as set E. However, it is assumed that a plurality of states Sjk of one Ij do not exist simultaneously in one set E. For example, if I1 takes state S11, state S12, state S13, and I2 takes state S21, state S22, state S23, state S24, the set E = {S11, S21} can be taken, but the set E = {S11, S12} cannot be taken. This is because one interpretation finding or clinical information takes only one state. In the following description, the diagnosis name is described using the symbol D. In the present embodiment, the diagnosis name is assumed to be three values: primary lung cancer (denoted as D1), lung metastasis of cancer (denoted as D2), and others (denoted as D3). Further, the inference probability of the diagnosis name Dr (r = 1, 2, 3) when the set E is given as input information is expressed as P (Dr | E). Similarly, a subset of the set E is expressed as Ex, and the inference probability of the diagnosis name Dr when the subset Ex is given is expressed as P (Dr | Ex). In addition, the influence degree of Ex on the diagnosis name Dr is expressed as I (Dr | Ex).

  In step S3000, the input information acquisition unit 102 acquires input information regarding an abnormal shadow of the lungs input to the diagnosis support apparatus 100 and data associated therewith. For example, in the acquired input information, I1 “shape”: state S12 “branch”, I2 “cut”: state S21 “strong”,..., Il “roll (bronchi)”: state Sl1 “present”,. “Past”: State Sm2 is “None”. In this case, the set E (that is, input information) of the state Sjk is E = {S12, S21,..., Sl1,.

  In step S3010, the inference unit 104 infers a probability (inference result) that each abnormal shadow is a diagnosis name based on the input information (that is, E) acquired in step S3000. As an inference means at this time, various existing inference methods such as a Bayesian network, a neural network, and a support vector machine can be used. In this embodiment, a Bayesian network is used as an inference means. A Bayesian network is an inference model that uses conditional probabilities, and it is possible to obtain inference probabilities for each diagnosis name (probability that the case is each diagnosis name, also called posterior probability) when input information is input. is there. In the present embodiment, inference probabilities for the diagnostic names D1, D2, and D3 of the abnormal shadow are acquired.

  In step S3020, the impact level acquisition unit 106 acquires the impact level using the input information acquired in step S3000 and the inference result acquired in step S3010. Specifically, the subset Ex (that is, the subset of input information) of at least one set E (that is, input information) is acquired, and the degree of influence indicating the degree of influence on each diagnosis name is acquired for each subset Ex. . For example, in the case of E = {S12, S21,..., Sl1,..., Sm2} in the above example, when acquiring a subset Ex having two elements, {S12, S21}, {S12, Sl1}, Ex such as {S21, Sm2} is acquired. In the present embodiment, it is assumed that all subsets Ex having one element are acquired. Therefore, in the above example, since there are m elements of E, m subsets Ex are acquired.

Next, the degree of influence is acquired in each subset Ex. In the present embodiment, the degree of influence is calculated using the prior probability of each diagnosis name and the inference probability when only Ex is input. In the present embodiment, the prior probability of each diagnosis name corresponds to the probability of each diagnosis name when there is no input. Then, for example, the Ex influence degree I (Dr | Ex) on the diagnosis name Dr is calculated by the following equation.

  In Formula (1), if I (Dr | Ex) is positive, that is, if the posterior probability when only Ex is input is greater than the prior probability, it is determined that Ex has a positive effect on Dr. On the other hand, if it is negative, that is, if the posterior probability when only Ex is input is smaller than the prior probability, it is determined that Ex has a negative effect on Dr. The influence degree calculation method described above is an example of processing in the present embodiment, and is not limited to this.

  In step S3030, the selection unit 108 selects a subset based on the inference result acquired in step S3010 and the degree of influence acquired in step S3020. In the present embodiment, the selection unit 108 selects a subset according to the inference probability and the threshold value of the diagnosis name having the highest inference probability among the inference results. Specifically, when the highest inference probability is equal to or greater than a predetermined threshold, the selection unit 108 selects a subset that affirms the diagnosis name having the highest inference probability. On the other hand, when the highest inference probability does not exceed the predetermined threshold, the selection unit 108 determines that the inference probability is low, and selects a subset that negates the diagnosis name other than the diagnosis name having the highest inference probability. Note that the above selection method is an example of processing in the present embodiment, and is not limited to this.

  The processing according to the present embodiment when the threshold is set to 70.0% will be described in detail with reference to FIG. FIGS. 5A and 5B show the inference probability P (Dr | Ex) obtained by the process of step S3010 and the process of step S3020 when different input information (set E) is acquired. The degree of influence I (Dr | Ex) on each diagnosis name obtained is shown. In FIG. 5A, the diagnosis name having the highest inference probability is D1 (ie, primary lung cancer), and the inference probability P (D1 | E) is 47.6%. In this case, since the inference probability P (D1 | E) does not exceed 70.0% of the threshold value, the selection unit 108 denies D2 and D3 other than D1, that is, I (D2 | Ex), I (D3 Select a subset for which | Ex) is negative. In the present embodiment, the selection unit 108 selects Ex that takes the smallest value from among the subsets Ex in which I (D2 | Ex) and I (D3 | Ex) are negative. Specifically, in the present embodiment, the selection unit 108 selects {Sm2} (= -13.8%) and {S43} (= -8.80), respectively.

  On the other hand, in FIG. 5B, the diagnosis name having the highest inference probability is D2 (that is, lung metastasis of cancer), and the inference probability P (D2 | E) is 82.1%. In this case, since the inference probability P (D2 | E) exceeds 70.0% of the threshold, the selection unit 108 selects information that affirms D2, that is, a subset in which I (D2 | Ex) is positive. In the present embodiment, the selection unit 108 selects Ex having the largest value among the subsets Ex in which I (D2 | Ex) is positive. Specifically, in the present embodiment, the selection unit 108 selects {S11} (= 10.4).

  In step S3040, the presentation unit 110 generates and displays information to be presented. Specifically, the information to be presented is generated based on the input information and accompanying data acquired in step S3000, the inference result acquired in step S3010, and the subset information selected in step S3030. FIG. 6 shows an example of the presentation information displayed on the monitor 1005 in the present embodiment when the inference result and the degree of influence shown in FIG. The presentation information 600 includes a representative image 6000 of abnormal lung shadows acquired in step S3000 and input information 6010 of abnormal lung shadows. In addition, the presentation information 600 includes the inference result 6020 inferred in step S3010. In the example shown in the figure, as the inference result 6020, an inference probability 6021 of primary lung cancer, an inference probability 6022 of cancer lung metastasis in the inference result, and other inference probabilities 6023 are displayed by a pie chart. In addition, the presentation information 600 includes a diagnosis name (primary lung cancer in the example in the figure) and its probability 6030 (in the example in the figure, the primary lung cancer of the cancer) in the inference result. Inference probability). In addition, the presentation information 600 includes support information 6040 selected by the process of step S6030.

  According to the present embodiment, the support information selection method is changed by considering the inference probability of the inference result. Specifically, when the inference probability of the target diagnosis name is high, information that affirms the diagnosis name is selected, and when it is low, information that denies another diagnosis name is selected. As a result, it is possible to present information similar to the case of interpretation by a doctor as support information. Therefore, the information required by the doctor can be presented efficiently. In other words, according to the present embodiment, since information according to the degree of confidence of the target diagnosis name is selected and presented, information required by the user can be presented efficiently.

[Modification 1]
In the present embodiment, in step S3030, information is selected according to the inference probability and threshold value of the diagnosis name having the highest inference probability. However, not only the inference probability of the highest diagnosis name but also the inference probabilities of other diagnosis names may be used as selection criteria. For example, information may be selected depending on whether the difference / ratio with the inference probability of the diagnosis name having the second highest inference probability exceeds a predetermined threshold. Alternatively, a difference / ratio with the inference probability of the diagnosis name having the lowest inference probability and a threshold value may be used. Further, the difference / ratio with the sum of the inference probabilities from the second to the Nth may be used, or the difference / ratio with the median of the inference probabilities of each diagnosis name may be used. According to this, since the selection of information is performed dynamically, the information required by the doctor can be flexibly presented.

[Modification 2]
In the present embodiment, in step S3030, information is selected according to the inference probability and threshold value of the diagnosis name having the highest inference probability. However, the inference probability of the diagnosis name having the second and subsequent inference probabilities may be used as a selection criterion. For example, information may be selected using the inference probability and threshold value of the diagnosis name having the lowest inference probability.

[Modification 3]
In the present embodiment, in step S3030, information is selected using the inference probability of the diagnostic name having the highest inference probability. However, instead of the diagnosis name having the highest inference probability, the inference probability of the diagnosis name specified by another method may be used. For example, information may be selected according to the inference probability of a diagnosis name designated by a doctor. According to this, since information is selected for a diagnosis name desired by the doctor, information required by the doctor can be efficiently presented.

[Modification 4]
In step S3030, when selecting information that denies the diagnosis name other than the target diagnosis name, the information is selected by simultaneously using the degree of influence of each diagnosis name other than the diagnosis name. However, the information may be selected using the influence degree of the diagnosis name as a target. Specifically, information that denies other diagnostic names while affirming the target diagnostic name may be selected. Moreover, you may select the information which denies multiple other than the diagnostic name used as object simultaneously. According to this, since information is selected using the degree of influence on a plurality of diagnosis names, information required by a doctor can be presented efficiently.

[Modification 5]
In step S3030, when the threshold value of the diagnostic name having the highest inference probability is exceeded, information that affirms the diagnostic name is presented, and when not exceeded, information that denies other diagnostic names is presented. . However, other methods may be used. Specifically, information that denies other diagnosis names may be presented when the diagnosis name threshold is exceeded, and information that affirms the diagnosis name may be presented when the diagnosis name threshold is not exceeded.
According to this method, since information is presented from the viewpoint opposite to the method normally performed by a doctor, it can be expected that the doctor uses the information as an alert when making a diagnosis.

  As described above, according to the present embodiment, the inference probability for each diagnosis name obtained from the interpretation findings and clinical information included in the medical information, and the influence of each of the findings and the subset of clinical information on the inference related to the inference probability. Appropriate support information can be provided based on the degree.

[Other Embodiments]
The present invention can also be realized by executing the following processing. That is, software (program) that realizes the functions of the above-described embodiments is supplied to a system or apparatus via a network or various storage media, and a computer (or CPU, MPU, etc.) of the system or apparatus reads the program. It is a process to be executed.

DESCRIPTION OF SYMBOLS 100 Diagnosis support apparatus, 102 Input information acquisition part, 104 Reasoning part, 106 Influence degree acquisition part, 108 Selection part, 110 Presentation part, 200 Information input terminal

  As a means for achieving the above object, the diagnosis support system of the present invention comprises the following arrangement. That is, an inference means for acquiring inference probabilities of a plurality of diagnosis names corresponding to the case based on a plurality of medical information corresponding to the case, and the inference probabilities of a target diagnosis name among the plurality of diagnosis names Based on the selected medical information, the selection means for selecting the medical information that negatively affects the inference probability of the diagnostic name other than the target diagnostic name when the is lower than a predetermined criterion Display control means for displaying support information on a display unit.

Claims (10)

Obtaining means for obtaining input information corresponding to a case;
An inference means for identifying a diagnosis name corresponding to the case based on the input information and obtaining an inference probability of the diagnosis name;
Display control means for displaying support information on the display unit according to the inference probability of the identified diagnosis name;
A diagnostic support system comprising: For the specified diagnosis name, further comprising an influence degree acquisition means for acquiring the influence degree of the input information,
The diagnosis support system according to claim 1, wherein the display control unit displays the support information based on the degree of influence.   The display control means displays information having an influence level that affirms the specified diagnosis name when an inference probability for the specified diagnosis name is higher than a predetermined criterion. The diagnosis support system described in 1.   When the display control means determines that the inference probability for the specified diagnosis name is lower than a predetermined reference, the display control means displays information having an influence level that negates a diagnosis name other than the specified diagnosis name. The diagnosis support system according to claim 2, wherein:   When the display control unit determines that the inference probability for the specified diagnosis name is lower than a predetermined criterion, the display control unit affirms the specified diagnosis name and diagnoses other than the specified diagnosis name The diagnosis support system according to claim 2, wherein information having an influence level of negating a name is displayed.   The diagnosis support system according to any one of claims 1 to 5, wherein the specified diagnosis name is a diagnosis name having the highest inference probability for the diagnosis name.   The diagnosis support system according to any one of claims 1 to 5, wherein the specified diagnosis name is specified by a user's designation.   The diagnosis support system according to claim 3, wherein the predetermined criterion is a threshold value. An acquisition process for acquiring input information corresponding to a case;
An inference step of identifying a diagnosis name corresponding to the case based on the input information and obtaining an inference probability of the diagnosis name;
A display control step for displaying support information on the display unit according to the inference probability of the specified diagnosis name;
A diagnostic support system control method comprising:   A program for causing a computer to execute each step of the diagnosis support method according to claim 9.

Images